MVALID (was Re: M-Label or MVALID, and dangers with mappings?)

John C Klensin klensin at jck.com
Tue Apr 14 18:25:19 CEST 2009



--On Tuesday, April 14, 2009 17:00 +0200 Patrik Fältström
<patrik at frobbit.se> wrote:

>...
> Mark, once again you try to talk about domain names, but do
> talk about URIs/IRIs, and you talk about HTML. That is not
> Domain Names, but one specific context.

I want to reinforce Patrik's comment because this is, IMO, very
important.  

While many of us tend to conflate them at one time or another,
there are three sets of issues here.  They are separable and
should be separated:

(1) Domain names.  Domain names are rarely used independent of a
context, but they are separate things.  IDNA recognizes this,
even in IDNA2003, by distinguishing between IDNA-aware and
IDNA-unaware contexts.  One cannot use a non-ASCII string in the
leading labels of an SRV records, one cannot use labels that
contain anything but European Digits in the ENUM tree, etc.
And, while the DNS can store and retrieve any string of octets,
IDNA does not recognize anything but Non-reserved LDH labels and
IDN Reserved LDH Labels (that is true even under IDNA2003
although it does not use that terminology).

(2) URI and IRI contexts.  These have different rules from
domain names generally and impose their own sets of constraints.
If we have a URI or IRI issue, then we should be talking about
that issue in URI or IRI terms (or the transition between them)
and not as part of domain names or IDNs generally.

(3) HTML, XML, and HTTP contexts.  These are partially a subset
of the URI/IRI issues, but introduce additional constraints and
issues partially because of a history of relaxed interpretations
of the existing protocols (even the HTML one).  Again, there are
legitimate questions and considerations about these contexts,
but it is unwise to assume that those issues are domain name
issues in general.

    john





More information about the Idna-update mailing list